Evaluating the aerodynamic design of moving objects, such as ordnance, is often done using high-speed imagery on film or videotape. The imagery must then be analyzed to determine the position and orientation of the object in each frame; from this data, the trajectory, acceleration, and other parameters of motion can be computed. In the past, the object's location was determined by human operators locating a few points or fiducial targets in each frame. A system for analyzing film images was developed based on a high resolution digital camera and a high performance image processing computer which, using a whole-object model, precisely locates objects in six degrees-of-freedom (6 DOF). Model-based image processing has been developed that substantially automates a manual three-dimensional (3-D) visualization process using two- dimensional (2-D), short focal length imagery. The model-based image processing algorithm has seven levels: (1) edge extraction; (2) edge modeling; (3) association of extracted edges with model components and measurement of displacement of edge pixels from them; (4) estimation of straight lines fit to each set of edge pixels; (5) qualification of extracted displacement equations based on self and mutual consistency; (6) estimation of object parameters and error statistics from one image; and (7) estimation of object parameters, trajectory, and error statistics from multiple viewpoints. Examples illustrate these steps of the image processing algorithms.
展开▼
